Time  Nick         Message
00:02 Exio4        there will be a 0.4.9 in xmas
00:07 sapier       did anyone ever try this one? minetest.hash_node_position?????
00:13 ShadowNinja  emptty: That's been around for a long time(Pre-0.4.3?), but usually it doesn't flash.
00:14 emptty       Exio4: cool! that's a good news
00:15 emptty       ShadowNinja: well, it was playable with 0.4.7, but it becomes rather difficult with the latest stable
00:15 emptty       (it was playable, even if it blinked from time to time, like when switching between "burning" and "not burning")
00:25 sapier       sadly hash_node_position is really a "hash" and not a 48 bit value
00:30 emptty       woot, it works perfectly with those three patches
00:30 emptty       thanks guys
00:40 ShadowNinja  sapier: return (pos.z+32768)*65536*65536 + (pos.y+32768)*65536 + pos.x+32768
00:41 ShadowNinja  sapier: Looks like a number to me...
00:41 sapier       if you do it backwards you see loss of z coordinate
00:43 sapier       funny thing is it's that symetric it can even be fixed
00:47 sapier       ok I don't understand whyt this code works https://gist.github.com/sapier/7865813
00:48 sapier       ohhhh
00:49 sapier       guess I should've copied it  to gist previously :-)
01:03 iqualfragile sapier: what exactly do you not understand about it?
01:05 sapier       I didn't see the typo in 65535 ;-)
01:06 iqualfragile k
01:14 leo_rockway  greetings, y'all.
01:14 leo_rockway  the last few days I've been getting segfaults with latest builds.
01:14 leo_rockway  I just figured out it can be reliable reproduced with the indev map generator
01:15 leo_rockway  I start minetest, create a new map using indev and when I try to run it it segfaults.
01:16 kaeza        ^ confirmed
01:20 leo_rockway  kaeza: ^ unconfirmed now? >.<
01:20 kaeza        wait...
01:23 kaeza        what is the point of the mapgen combo on the map settings if it's only applied when you actually enter the map (i.e. not after creating it) ?
01:23 kaeza        anyway, seems to crash reliably on indev
01:24 leo_rockway  alright
01:24 kaeza        for now, this is the gdb bt (release build): http://pastebin.com/pcWUJagb I'll build a debug version
01:25 VanessaE     I thought we weren't supposed to use release builds for now?
01:26 VanessaE     was that already fixed?
01:34 kaeza        http://pastebin.com/ma3kGpHt
01:46 daswort      i get this error over and over again. looks a bit messy. https://dpaste.de/5LWx
01:46 daswort      why does MT write that error thousand times in the output even when there is not interaction with the menu?
01:50 sapier       yes VanessaE
01:50 sapier       got fixed following day
01:50 VanessaE     oh ok.
01:51 VanessaE     as always I'm a day late and a dollar short :P
01:51 sapier       indev is not recommended for production use
01:51 VanessaE     looks like it's not recommeded for *any* use at the moment :)
01:51 sapier       but be aware release builds usually are more crashy then debug builds
01:52 sapier       master is quite buggy atm ... guess we need to stop adding experimental things if we want to release until christmas
01:53 sapier       a green tree ... I still don't have a christmas tree ... <<< fetching saw >>>
01:56 VanessaE     ha!
02:57 hmmmm        I see the problem with proller's mapgen
02:57 hmmmm        he feels that he doesn't need to set ndef inside of his cave's ctor override
08:13 RealBadAngel nore's patch broke shaders on some GPUs, heres the fix (and some optimization, thx to nore's idea)
08:13 RealBadAngel https://github.com/RealBadAngel/minetest/commit/39497c05b3ddebaec1783a16b6cec612d4b66ff7
08:13 RealBadAngel any objections?
08:14 RealBadAngel btw patch is already tested by me, nore and VanessaE
08:15 VanessaE     works fine for me, no regressions that I could find after significant testing.
08:15 nore         I'm ok for it...
09:18 specing      man it would be great if minetest actually remembered passwords
09:18 VanessaE     G*d yes
09:18 specing      It would be much more great if it wre using pubkey auth!
09:18 specing      (quake style key)
09:56 specing      I think boats are actually slower than jsut swimming
16:52 sapier       https://github.com/minetest/minetest/pull/1030 come on guys that one is sitting around for days now fully tested lacking ANY comment of a core dev
16:54 sapier       same for this one https://github.com/minetest/minetest/pull/1027
16:54 sapier       https://github.com/minetest/minetest/pull/1036 and of course that one too ... john_minetest did quite a lot of testing for those
16:55 thexyz       I think that for #1027 we should have a one hardcoded (client-side setting) CDN server which the client will always use before attempting any others
16:55 thexyz       this will improve player experience on servers that don't have remote media set up
16:56 sapier       interesting idea but wouldn't that put a lot of load to that server?
16:57 thexyz       for now we have minetest-cdn.net which is cloudflare'd
16:57 thexyz       and i set it up so that media requests are served by cloudflare
16:57 sapier       what about making server to tell client where to look for data? this way it's responsibility of server owner
16:58 thexyz       and the server itself only sets a redirect
16:58 thexyz       it's already this way, isn't it?
16:58 thexyz       the problem is that on old servers no one is going to care about this
16:58 thexyz       and having a "default" cdn server just works
16:58 sapier       thought it's always fetched from server
16:59 thexyz       what do you mean?
16:59 sapier       yes but how is client supposed to know textures on cdn are same server has?
16:59 sapier       e.g. if at server default textures are replaced
17:00 thexyz       did you even read this pull you're suggesting to merge?
17:00 thexyz       it uses sha hashes now instead of texture names
17:00 thexyz       and verifies them of course
17:01 sapier       I know so if server replaced the textures you do a useless lookup at cdn
17:02 sapier       I read it quite some time agon ... it's sitting around for months now ... can't tell I know any singel detail about it
17:10 thexyz       sapier: what do you mean "replaced the textures"?
17:10 thexyz       it works this way: client gets list of hashes from server, POSTs them to the CDN server, gets list of hashes of textures the CDN server has and requests them from the CDN server
17:11 thexyz       then with remaining textures/those that it failed to fetch it switches to the next CDN server
17:11 thexyz       and in the end just ask server to send them the usual way
17:15 sapier       this way EVERYTHING is checked by cdn server even if gameserver already KNOWS cdn server can't have the texture
17:16 thexyz       gameserver doesn't know anything about cdn
17:16 sapier       and btw I have a very very bad feeling about minetest calling 100 servers once it's started without user knowing anything about it
17:16 thexyz       it doesn't call 100 servers either
17:16 thexyz       everything is checked, yes, but that's not the problem
17:16 BlockMen     a simple question. does another one need to agree for a typo fix in lua-api, or is it fine just to push
17:17 thexyz       (hint: calculate size of the hash we're using, multiply by average number of textures and compare to some texture .png)
17:17 sapier       every 2 months another server is added to be autochecked
17:17 sapier       serverlist
17:17 thexyz       what?
17:17 sapier       modlist
17:17 sapier       cdn server
17:17 thexyz       as you wish..
17:17 sapier       what's going to be next? nsa "I'm now playing minetest" server
17:17 thexyz       I just felt that it'll improve player's game experience greatly
17:17 thexyz       no more lags even when playing on old servers which aren't going to be updated
17:18 thexyz       (when receiving media, that is)
17:18 sapier       it may help yes but I don't like it to be default
17:18 thexyz       whatever
17:18 thexyz       you should ask kahrl if his branch is merge-ready though
17:18 sapier       if it's disabled by default and you have to enable it in settings it's fine
17:19 thexyz       no it's not
17:19 thexyz       because no one will do this and those who won't do this (read: everyone) will complain about media transfer being very slow
17:19 sapier       it is as it's an active decision of user so he's aware of it
17:19 thexyz       and no one ever will complain about requests being made to some public cdn server
17:20 sapier       usually if commercial software does callbacks to servers I don't know I instantly block access to those servers
17:21 sapier       noone complains about nsa catching all data ... because most people can't imagine what amount of data is collected
17:21 thexyz       now that's some shitposting
17:22 thexyz       so I guess this conversation is pointless
17:22 thexyz       just ask kahrl if his branch is ready and if it is then just throw it in
17:22 sapier       no it isn't user does connect to a server whos ip or name he entered
17:22 sapier       usually I don't expect a application to call other servers too
17:22 thexyz       this won't lead us anywhere
17:22 thexyz       and I'm not in the mood to argue about it; I'll just fix it in the fork
17:23 sapier       quite obvious ... guess we need to agree that we don't have same opinion only
17:23 thexyz       indeed
17:23 celeron55    thexyz: you shouldn't argue about it endlessly with a single other core dev; it's pointless
17:23 thexyz       I know
17:23 thexyz       rather than that, I shouldn't argue
17:24 sapier       how much forks are out there right now?
17:24 thexyz       oh
17:24 thexyz       I thought about creating a "minetest family tree" or something like that
17:24 celeron55    i don't have much against using minetest-cdn.net as the default; but i know some people will hate it
17:24 thexyz       but I don't know minetest history enough
17:24 thexyz       I know about minetest-classic, freeminer; all others seem to be dead
17:25 celeron55    there are one or two (or three?) forks by jachoo
17:25 thexyz       aren't those dead?
17:25 sapier       if people do forks to get features instead of creating a game I guess we're doing something wrong
17:25 celeron55    yes, but they would be interesting to have there
17:25 thexyz       sapier: or maybe right?
17:25 thexyz       who knows
17:26 celeron55    sapier: the only thing it means is that people have different opinions
17:26 sapier       maybe but our intention was to create a engine as flexible as possible ... imho that means forkin wouldn't be necessary
17:26 celeron55    anyway, i also once made a fork of 0.4 for a paid project where some guy wanted to explore and dig around in his custom-generated voxel worlds 8)
17:27 celeron55    and then bought a laptop with that money
17:28 sapier       I'd not call this a "real fork" :-) but yes as of git naming it is
17:28 celeron55    i consider dead forks very important
17:28 celeron55    there's always a lot to learn from in them
17:29 celeron55    because they always become to exist for a reason
20:29 specing      /data/specing/programs/minetest/minetest/src/inventory.cpp:624: ItemStack& InventoryList::getItem(irr::u32): Assertion 'i < m_size' failed.
20:29 specing      In thread 7f444d5867c0:
20:30 sapier       what did you do?
20:30 sapier       what mt version?
20:30 specing      http://bpaste.net/show/LcHYEROz2Uy5tc7AE7C7/
20:30 specing      sapier: right click a craft guide
20:30 specing      4.8.0-dev (updated saturday?)
20:30 PilzAdam     whats the server version?
20:31 specing      whatever clip's public server is running
20:31 VanessaE     0.4.8-dev
20:32 specing      http://bpaste.net/show/o77onbC1u0Gj5zG7isMW/
20:32 specing      full session
20:32 specing      I clicked a craft guide near spawn
20:33 PilzAdam     does it only happen for this single craftguide?
20:34 specing      I think it is pretty reproducable
20:34 specing      I got atleast 5 crashes on saturday
20:34 specing      not sure if only this one
20:35 sapier       seems we still miss a case of invaid data passed to scriptapi
20:35 PilzAdam     yea, I guess its a mod that messes it up
20:35 sapier       still we shouldn't crash that way
20:36 sapier       specing can you try to find out what is required to reproduce it? or maybe run within gdb?
20:37 PilzAdam     whats the server's IP?
20:41 specing      sapier: click on that craft guide :/
20:41 specing      jsut crashed again
20:41 PilzAdam     sapier, https://gist.github.com/PilzAdam/7880453
20:41 specing      yes, I saw PilzAdam crash :D
20:43 Exio4        >optimized out
20:44 sapier       argh ... no idea why inventorylist believes to have to hande a size different from the one in its internal vector
20:46 sapier       ok inventory list is a little bit more complex to understand what's going on in 30 seconds
20:47 PilzAdam     specing, how does the craftguide formspec look when it doesnt crash?
20:48 PilzAdam     or better, where is the mod sourcecode?
20:52 sapier       InventoryList uses inconsistent size that's not a mod error
20:52 sapier       inventory.cpp L592 returns size of vector direct
20:53 sapier       while assertion L618 and L624 is against m_size
20:53 sapier       and for delete m_items.size() is used again
20:56 specing      PilzAdam: it looks like a crafting guide and idk where the source is
20:57 PilzAdam     sapier, and where do they get out of sync?
20:57 PilzAdam     also changing m_size to m_items.size() in the assert() doesnt fix the crash
20:57 sapier       that'S what I haven't figured out but obviously they do
21:04 VanessaE     guys, speaking of crashes, how long are you going to let the chat crash that nearly everyone gets bit by persist before investigating and fixing it?
21:05 specing      Can someone implement quake-like client-server auth?
21:05 PilzAdam     sapier, if(item_i >= (s32) ilist->getSize() || item_i < 0) guiFormspecMenu.cpp:1728 fixes it
21:06 sapier1      great so you do the right check and don't SEE the error any longer
21:07 sapier1      l_inventory.cpp checks for m_size while the called getItem() checks against vectorsize .... meaning in this special case m_size and actual vector size have to be out of sync
21:07 PilzAdam     <PilzAdam> also changing m_size to m_items.size() in the assert() doesnt fix the crash
21:07 PilzAdam     seems like the index is just negative
21:08 sapier1      yes changing return m_size to return m_items.size() will fix the issue too
21:09 sapier1      if I'm correct and really l_get_stack is called then index can't be negative
21:09 sapier1      ok let's double check that assumption first
21:10 sapier1      I'd have to put a breakpoint there but according to eclipse search that's the only location at all where getItem is called
21:26 sapier       (22:01:49) PilzAdam: also changing m_size to m_items.size() in the assert() doesnt fix the crash ooops ... where does that getItems() call originate from?
21:26 PilzAdam     https://gist.github.com/PilzAdam/7880453
21:28 sapier       590 lines of backtrace ;-/
21:29 PilzAdam     well, 530 of it are the_game + main()
21:29 sapier       ok I've been completely wrong ... sorry
21:33 PilzAdam     hah
21:33 PilzAdam     list[context;space;0,0;10,1;-2] causes the crash
21:33 PilzAdam     i.e. a negative starting index
21:33 sapier       already admited I was wrong ;-)
21:34 sapier       but that size handling should be cleaned too
21:34 PilzAdam     now where is the list formspec parsed?
21:34 sapier       GuiFormSpecMenu.cpp
21:36 Megaf        Folks, could you please fix this fucking annoying bug? This stinking shit keeps crashing when ones TRIES to send a not small chat message ~10 words
21:36 Megaf        http://paste.debian.net/70051/
21:36 Megaf        I'm totally fed up with this shit
21:36 sapier       any idea how to reproduce it?
21:37 Megaf        download whatever new build you want, stable or master branch
21:37 Megaf        compile it whatever you want too
21:37 sapier       doesn't relieably crash for me everytime I trie to get it it's rocksolid
21:37 Megaf        join to whatever server, and write whatever you want with + 10 words
21:37 sapier       but I've seen those crashes too ... just everytime when I don't use a debugger
21:38 PilzAdam     sapier, https://gist.github.com/PilzAdam/7881433 ?
21:38 Megaf        it will crash as soon as you hit enter and it wont send any message
21:38 sapier       ok push it
21:38 Megaf        also this bug is ~3 weeks old
21:39 Megaf        perhaps it happens only in my server anyhow
21:40 Megaf        its megaf.co.vu port 30003
21:43 PilzAdam     specing, its fixed now, get the latest client
21:44 PilzAdam     oh, and whoever wrote the craft guide mod needs to fix it
21:46 sapier       ok megaf I stop annoying your users doesn't seem to depend on your server
21:47 Megaf        Ok, thanks
21:47 Megaf        glad that it's not an issue with my server